fig-cap-location conflict with code annotation

[cderv]

---
title: 'Title'
format: html
---

::: {#fig-learnxmp fig-cap-location="top"}
```xml
<context-info>
  <url>file:/Volumes/.../1051-Learn.mp4</url>           # <1>
```
1. Testing annotations.

Caption

:::

while without annotation

---
title: 'Title'
format: html
---

::: {#fig-learnxmp fig-cap-location="top"}
```xml
<context-info>
  <url>file:/Volumes/.../1051-Learn.mp4</url>           
```

Caption

:::

Same with Margin

Found while looking at https://forum.posit.co/t/code-annotations-after-caption-in-labeled-container-divs/183061

[cscheid]

This is going to be really hard to fix...

[cderv]

Ok - so in the mean time, we may want to document the limitation or add a warning (if possible) about this conflict.

Not the first conflict between features that we have (which is totally ok to be honest), so it could be worth having a mechanism to warn about such things, or at least a way to document it nicely.

Anyhow, I'll put that in Future for now then.

[cscheid]

Not the first conflict between features that we have (which is totally ok to be honest), so it could be worth having a mechanism to warn about such things, or at least a way to document it nicely.

Yes, 100% agreed.

[cscheid]

I'm looking into this, and what's happening specifically in this example is that Quarto is getting confused about the code block being the content of a figure, and lifting it out of the float. This is a real bug, but I'm not sure how we can stop it from happening.

If you replace fig- with lst-, you'll see the problem goes away.

But in order to know which code blocks are "content" and which code blocks are the echo of a code cell, I think we need additional information that isn't in the document right now.